Type | Workflows |
Stats | 734 |
Reviews | (42) |
Published | Jun 26, 2024 |
Base Model | |
Hash | AutoV2 7458F86ACA |
You only need one photo of the face you want to replace and one reference image to generate your desired AI photography artwork.
Your required inputs include:
A high resolution reference image of the person's face. (2K resolution with majority of picture is face, lower quality might result in blurry face in generation image, but usually quality would be fine)
A reference image for the desired photo.
Positive prompt: Simply describe the main content of the reference image, leaving more room for extension in generation of image. (If your reference image is one you generated yourself, you can use the existing prompts. If it's another image, you can briefly describe the content and colors. No need to use any LoRAs or describe the style, as most of the work will be handled by Tile Reference and IPAdapter.)
Negative prompt: Improve the quality of the generated image with negative prompt words, such as "blurry" and "worst quality".
Extensions you need to install the following, referring to the GitHub pages for installation and downloading the necessary models:
IPAdapter-plus: https://github.com/cubiq/ComfyUI_IPAdapter_plus
Provides style reference control.
InstantID: https://github.com/cubiq/ComfyUI_InstantID
Provides SDXL face-swapping.
Regthree: https://github.com/rgthree/rgthree-comfy
ComfyUI Essential: https://github.com/cubiq/ComfyUI_essentials
Both are excellent ComfyUI control plugins.
ComfyUI Advanced ControlNet: https://github.com/Kosinkadink/ComfyUI-Advanced-ControlNet
For advanced ControlNet use in ComfyUI, you can use your own ControlNet plugins.
TTPlanet Tile Reference Model: https://civitai.com/models/330313/ttplanetsdxlcontrolnettilerealistic
Image preview chooser: https://github.com/chrisgoringe/cg-image-picker
You can generate 4 imgs then select one before the refine stage and high-resolution fix to save little time.
The intention behind this is that many people are making money by selling courses on so called AI photography, but the aesthetic and quality of their output are often poor and techniques are simple. I created a simplified workflow so that everyone can have the AI photography images they want. This workflow can also be used to generate your own cartoon avatars and more, with very high quality. I used "leosam Helloworld" as base model in all these images generation.
Leave a message if you got a any problem.
Reference image examples: (First one is a real photography, 2nd and 3rd are images I generated through MJ)
Face embedding samples:
Result samples:
Example positive prompt:
Surrealist photography, 1 female futuristic electronic devices white ceramic gown, classical Gothic church, looking at the viewer, rich in detail, high-definition sharp focus, large aperture, blurred background
Example positive prompt:
A woman in green dress sitting on sofa, crossed leg, earring, high resolution, 4k, 8k,
Example positive prompt:
long black hair, fashion photography, 1 female beautiful lady posing, looking at the viewer, wearing a insanely detailed sparkling silver dress, in a runway, pale skin, sharp focus large aperture, blurred background, science fiction, rich in details, low saturation, high resolution, 4k, 8k
